Re: Mounting EBS volumes

From: Ryan G <no email>
Date: Sat, 4 Apr 2015 14:15:39 -0400

Nevermind. I see this is actually 'DEVICE' in cluster.py except it only
likes /dev/sd*. I think this should be changed to allow /dev/xvda*. In
CentOS 6, /dev/sd* gets renamed. I've discovered if I specify an EBS
volume device path to /dev/xvd*, it holds.

On Sat, Apr 4, 2015 at 2:02 PM, Ryan G <ngsbioinformatics_at_gmail.com> wrote:

> I'm playing around with using StarCluster and different AMIs. One of the
> things I've noticed is that EBS volumes don't get mounted where expected.
> For instance, my config file specifies something like:
>
> [volume biodata]
> # attach vol-c9999999 to /home on master node and NFS-shre to worker nodes
> VOLUME_ID = vol-c999999
> MOUNT_PATH = /home
>
> StarCluster then tries to attach the volume at /dev/sdz, But ends up
> giving the error "Cannot find device /dev/xvdz". When I log in, I see the
> volume got attached to /dev/xvdaz.
>
> I can either modified my copy of StarCluster to correct for this, but I
> think a better solution would be to be able to specified the block device:
>
> [volume biodata]
> # attach vol-c9999999 to /home on master node and NFS-shre to worker nodes
> VOLUME_ID = vol-c999999
> MOUNT_PATH = /home
> BLOCK_DEVICE = /dev/xvdaz
>
> This way, the user can control where the mount is happening and expected.
>
> Thoughts?
>
Received on Sat Apr 04 2015 - 14:15:41 EDT